Table 1.

Characteristics of studies included in a meta-analysis of trials assessing the effectiveness of lithium for prevention of mood episodes in bipolar disorders

Study	Year	Comparator	Design	Participants	Interventions (with levels)	Definition of relapse/recurrence	Quality (rating)	Previous lithium/comparator use	Lithium/comparator serum level achieved
Prien 1973	1973	Placebo	Random assignment, 2-year follow-up	Patients with manic depressive disorder, manic type (n = 205), age 17 to 60 years; most recent episode: manic	Lithium (0.5 to 1.4 mEq/l); placebo	Emergent manic or depressive attack measured on Global Affective Scale requiring hospitalization (severe relapse) or supplementary medication (moderate relapse); combined moderate and severe relapse rates used	Allocation concealment unclear (B); participants and clinical raters blinded to treatment allocation; treating physician not blinded (A)	Following remission of the acute manic episode and prior to discharge (time of randomization) patients were stabilized on lithium (0.5 to 1.4 mEq/l)	Median serum lithium level 0.7 mEq/l
Kane 1982	1982	Placebo	Random assignment, up to 2-year follow-up	Patients with bipolar II disorder (Research Diagnostic Criteria) (n = 22), age 18 to 65 years; patients had been euthymic for 6 months prior to entry into the study	Lithium (0.8 to 1.2 mEq/l); imipramine (100 to 150 mg per day); lithium plus imipramine; placebo	Emergent mood episode meeting Research Diagnostic Criteria for major depressive disorder for 1 week, minor depressive disorder for 4 weeks, manic episode for any duration, or hypomanic episode for 1 week	Allocation concealment unclear (B); patients and physicians blinded to treatment allocation (A)	Patients had been on open uncontrolled continuation treatment for 6 months (except for the last 6 weeks (open treatment with imipramine)) before they were randomly assigned to treatment condition	Not available
Greil 1997	1997	Carbamazepine	Random assignment, 2.5 years observation period, primary aim was to assess efficacy of carbamazepine	Patients with current episode of bipolar affective disorder (ICD-9: 296.2, 296.3, 296.4) (n = 144), no preventive treatment immediately before current episode, age 18 to 65 years	Lithium (0.6 to 0.8 mmol/l); carbamazepine (4 to 12 μg/ml)	Recurrence, i.e. rating of psychopathology of 5 (=recurrence) or 6 (=extremely severe recurrence) corresponding to the recurrence of an affective episode in line with the Research Diagnostic Criteria	Non-blind design, randomization procedure by Efron (1971) (A); Allocation concealment adequate (A): central allocation through coordinating study centre, treatment group allocation by phone at the moment of randomization	Stabilization phase: psychotropic medication according to the free decision of the treating physician was gradually reduced and, if possible, discontinued before randomization; 84% never had received prophylactic treatment before	0.63 ± 0.12 mmol/l
Bowden 2000	2000	Placebo, valproate	Random assignment, 1-year follow-up, primary aim was to assess efficacy of divalproex	Patients with bipolar I disorder (DSM-III-R) with index manic episode according to Structured Clinical Interview for DSM-III-R (n=372), those with high suicide risk excluded, age 18-75 years	Lithium (0.8 to 1.2 mEq/l); divalproex (71 to 125 ug/ml); placebo	Emergent manic episode (Mania Rating Scale score of 16 or more or requiring hospitalization) or depressive episode (requiring antidepressant use or premature study withdrawal)	Allocation concealment unclear (B); patients, clinicians, and outcome assessors blinded to treatment allocation (A)	Before randomization, 117/372 were treated with open-label divalproex, 124 with lithium, 50 with both drugs, 81 with neither drug; lithium as well as divalproex were gradually reduced and withdrawn during the first 2 weeks of maintenance treatment	Mean (SD) serum lithium concentration by day 30: 1.0 ± 0.48 mEq/l; mean (SD) valproate concentration by day 30: 84.8 ± 29.9 ug/ml
Bowden 2003	2003	Placebo, lamotrigine	Random assignment, 1-year follow-up, primary aim was to assess efficacy of lamotrigine	Patients with bipolar I disorder recently recovered from a manic or hypomanic episode (DSM-IV) (n = 175), age ≥18 years	Lithium (0.8 to 1.1 mEq/l); placebo; lamotrigine (100 to 400 mg per day)	Intervention (additional medication or ECT) required for any mood episode; secondary outcomes subdivided by type of mood episode (manic/hypomanic/mixed or depressive)	Allocation concealment unclear (B); patients, clinicians, and outcome assessors blinded to treatment allocation (A)	Majority of participants had a prior history of lithium use (31/46: 70% in lithium group; 42/69: 67% in placebo group; 38/58: 72% in lamotrigine group). 18% of participants during the initial part of the 8- to 16-week open-label phase received lithium, the dosage of which was tapered over at least 3 weeks and discontinued a minimum of 1 week before they entered the double-blind phase of the study. All participants received open-label lamotrigine during the open-label phase (target dosage 200 mg/d; minimum 100 mg/d). Concomitant psychotropic medications were permitted during the open-label phase as needed to treat an ongoing manic or hypomanic episode but were discontinued a minimum of 1 to 2 weeks before entry into the double-blind phase.	Not available
Calabrese 2003	2003	Placebo, lamotrigine	Random assignment, 1-year follow-up, primary aim was to assess efficacy of lamotrigine	Patients with bipolar I disorder recently recovered from a major depressive episode according to DSM-IV (n = 463), age ≥18 years	Lithium (0.8 to 1.1 mEq/l); placebo; lamotrigine (50 to 400 mg per day)	Intervention (additional medication or ECT) required for any mood episode; secondary outcomes subdivided by type of mood episode (manic/hypomanic/mixed or depressive)	Allocation concealment unclear (B); patients, clinicians, and outcome assessors blinded to treatment allocation (A)	Majority of participants had a prior history of lithium use (57% to 62% of patients had received prior lithium treatment at some point, with 67% to 72% of these patients having achieved good clinical response and 80% to 85% having tolerated such prior treatment). 20% of participants in open label run in received lithium, dosage tapered over at least 3 weeks and discontinued a minimum of 1 week prior to entering the double-blind phase; any psychotropic medication permitted during 8- to 16- week open-label phase; all patients received lamotrigine (target dosage 200 mg/d; minimum 100 mg/d) as adjunctive therapy or monotherapy; all psychotropic medication other than lamotrigine were discontinued at least 7 days prior to randomization	Steady-state mean ± SD serum levels of 0.8 ± 0.3 mEq/l
Hartong 2003	2003	Carbamazepine	Random assignment, 2-year study	Patients with bipolar disorder (DSM-III-R) with at least two episodes during the last 3 years, recovered from last episode (n = 94), age ≥18 years	Lithium (0.6 to 1.0 mmol/l); carbamazepine (6 to 10 mg/l))	Recurrence of an episode of (hypo)mania or major depression according to DSM-III-R criteria	Allocation concealment adequate: pharmacy-controlled block randomization (A); double dummy design, double blind (A)	Total lithium/carbamazepine treatment during lifetime ≤6 months; at randomization, no patient received antidepressants, antipsychotics, or benzodiazepines.	Lithium level mean (SD): 0.75 (0.18) mmol/l; carbamazepine level mean (SD): 6.8 (1.2) mg/l
Geddes 2010	2010	Valproate	Random assignment, 24-month follow-up, primary aim was to assess efficacy of lithium-valproate combination therapy	Patients with bipolar I disorder on the basis of a previous episode of mania meeting DSM-IV criteria (n = 330), age ≥16 years; most recent episodes 52% mania, 34% depression, 12% mixed, 3% cycling	Lithium (0.4 to 1.0 mmol/l); valproate (750 to 1,250 mg)	Initiation of new intervention for an emergent mood episode, including drug treatment or admission to hospital	Randomization computerized, minimization; allocation concealment adequate: central allocation via telephone (A); investigators and participants informed of treatment allocation, trial management team masked to treatment assignment (A)	Before randomization active run-in of 4 to 8 weeks: all patients received lithium and valproate (lithium serum level 0.4 to 1.0 mmol/l; valproate dose at least 750 mg or valproic acid serum concentration at least 50 μg/ml)	Not available
Licht 2010	2010	Lamotrigine	Random assignment, up to 5.8-year follow-up, primary aim was to assess efficacy of lamotrigine	Patients with bipolar I disorder according to DSM-IV with at least two episodes within the last 5 years (n = 155) recruited during or in the aftermath of an index episode, age ≥18 years; index episode either depression (51%), mania (41%), or mixed mania (8%) according to the Cincinnati criteria, onset within the last year prior to randomization	Lithium (0.5 to 1.0 mmol/l); lamotrigine (up-titrated to 400 mg/day)	Psychotropic treatment (in addition to study drug and benzodiazepines) and/or hospitalization still required at month 6 after randomization; psychotropic treatment (in addition to study drug and benzodiazepines) during at least 1 week and/or hospitalization during at least 1 week still required after month 6 (after randomization)	Allocation concealment adequate: central allocation; computer-generated randomization plan, block randomization (A)	Prior lithium prophylaxis: 13 (17%) in lamotrigine group, 15 (19%) in lithium group; patients receiving lithium until randomization and assigned to lamotrigine group: lithium was tapered off over 1 to 3 months; patients receiving lamotrigine until randomization and assigned to lithium group: discontinuation of lamotrigine at randomization. Additional antipsychotic or antidepressant drugs were allowed in the first 6 months after randomization, investigators were encouraged to achieve monotherapy at month 6. Benzodiazepines allowed throughout the study.	Serum lithium level: mean 0.69 mmol/l (SD = 0.20); lamotrigine dose: mean 379 mg (SD = 66) (serum level 22.5 (12.7) μmol/l)
Amsterdam 2010	2010	Placebo, fluoxetine	Random assignment, up to 1-year follow-up	Patients with bipolar II disorder (n = 81) recently recovered from depressive episode with fluoxetine treatment, age 19 to 67 years	Lithium (0.5 to 1.5 mEq/l); fluoxetine (10 to 40 mg per day); placebo	Depressive relapse defined as HAMD score of 14 or more and meeting diagnostic criteria for major depressive episode. Hypomanic episode defined by DSM-IV criteria	Allocation concealment unclear (B); patients, clinicians, and outcome assessors blinded to treatment allocation (A)	Initial fluoxetine monotherapy was administered on the basis of response and tolerability. Patients who had a final HAM-D score ≤ 8 by week 12 of treatment were randomly assigned to different treatment arms. Patients assigned to fluoxetine group who previously took >40 mg/day of fluoxetine: dosage reduced to 40 mg/day; previously ≤ 40 mg/day: dosage maintained; patients assigned to lithium group: fluoxetine therapy discontinued. Lithium therapy initiated at 600 mg/day for 1 week, increased to 900 mg/day in week 2, continued until serum level of 0.5 to 1.5 mEq/l achieved by week 4.	Mean average serum lithium level was 0.69 mmol/liter (SD = 0.27), mean average maximum fluoxetine dose 34.3 mg/day (SD = 7.9)
Weisler et al. 2011	2011	Placebo, quetiapine	Random assignment, up to 2-year follow-up	Patients with bipolar I disorder (DSM-IV) recently recovered from a manic (53.6%), depressive (28%) or mixed episode (18.4%) (n = 1,172), age ≥18 years	Lithium (0.6 to 1.2 mEq/l); quetiapine (300 to 800 mg per day); placebo	Emergent mood event requiring medication or hospitalization, YMRS or MADRS 20 or more on two consecutive assessments, discontinuation attributed to mood event by investigator	Allocation by centralized randomization and drug allocation system (A); patients, clinicians, and outcome assessors blinded to treatment allocation (A)	All patients received open-label quetiapine (300 - 800 mg/d) for 4-24 weeks. Patients achieving stabilization on quetiapine were randomized to different treatment arms. Replacement of quetiapine tablets used during prerandomization phase started on day 1 and was completed by 2 weeks. Known intolerance or lack of response to lithium was an exclusion criterion	Mean (SD) median serum concentration was 0.63 (0.45) mEq/l; mean (SD) median quetiapine dose 546 (173) mg